1. Field of the Invention
The present invention relates to management technology of a quality of service provided by a distributed processing system.
2. Description of the Related Art
In a conventional distributed processing system, to economize on computational resources, an operation of a server for distributed object is stopped while the server is not in use and is then re-started when a call is issued by a client. A mechanism to carry out the activation of the server is called an xe2x80x9cactivation mechanismxe2x80x9d.
FIG. 2 is a schematic diagram showing the distributed processing system having the activation mechanism. The distributed processing system is mainly composed of the activation mechanism 3 used to allow communications between the client 1 and the server 2 and of a virtual reference section 4.
Procedures by which the activation mechanism 3 and the virtual reference section 4 are operated to activate the server 2 are hereinafter described.
First, in the distributed processing system, a calling request (i.e., a request for the call for operations of methods provided by the server 2) is given by the client 1 to the virtual reference section 4 (see reference number (1) in FIG. 2). When the virtual reference section 4 receives the above request, it checks if there is a real reference 4A internally and, if the real reference exists, the calling request is outputted to the real address space (i.e., an IP address and a port number) of the server 2 saved in the real reference section 4A. If the calling is made successfully, a reply is provided by the server 2, which will be forwarded to the client 1 by the virtual reference section 4.
However, if the call fails due to an absence of server 2 or an unexpected malfunction in server 2, the virtual reference section 4 outputs a request for activating a new server to the activation mechanism 3 so that the server 2 be activated (see (3) in FIG. 2). When the activation mechanism 3 receives the above request, it activates the server 2 in response to the request (see (4) in FIG. 2) and, at the same time, it gives a new real reference 4A storing the real address (i.e., an IP address and a port number) of the server 2 to the virtual reference section 4 (see (5) in FIG. 2). Thus, information required for activating the server 2 (i.e., location from which the server""s code will be downloaded, and/or the server""s persistent data) is managed by the activation mechanism 3.
When the virtual reference section 4 obtains the new real reference 4A, it, by using this new real reference 4A, outputs the calling requests again to the server 2 (see (6) in FIG. 2). If the call is made successfully, a reply is given from the server 2 to the virtual reference section 4 and the reply is then transferred to the client 1 (see (7)(8) in FIG. 2). After above operations, the client 1 can be in communication with the server 2.
Once the server 2 is activated, it remains activated until it stops for itself or it is instructed to stop by the activation mechanism 3. While the server 2 is in communication with the client 1, a communication path (for example, with an IP address and a port number of TCP/IP [Transmission control protocol/Internet protocol]) assigned at the time of the activation of the server 2 is used. The communication path is assigned by the activation mechanism 3 at the time of activating the server 2. The communication path can either be fixed to a pre-defined path (port number) or be assigned with a path (port number), which is dynamically allotted every time the server is activated.
As described above, the conventional distributed processing system has a problem in that the communication path is assigned statically or randomly regardless to the QoS control mechanism, which is utilized in the network nodes. Therefore, service quality of distributed objects cannot be controlled well.
In view of the above, it is an object of the present invention to provide a service quality management system which is capable of adaptively controlling the quality of services provided by a distributed object.
According to the first aspect of the present invention, a service quality management system is provided, which includes:
first storage means for saving relations between each of a plurality of servers for distributed objects and a priority defined for services provided by said each server;
second storage means for saving information about relations between an identifier of a communication path under the control of a node and a priority defined for processing performed by the node;
priority reading means for detecting a request for activation of a server and reading a priority defined for a service provided by a corresponding server from the first storage means;
identifier reading means for reading, from the second storage means, the identifier of the communication path corresponding to processing having a priority being equivalent to that read out from the priority reading means; and
server activating means for actually activating a corresponding server in accordance with the detected activation request and for connecting the server to a communication path designated by the identifier read by the identifier reading means.
In the foregoing, a preferable mode is one wherein, when a classification of levels of the priority given by the second storage means agrees with that of the priority given by the first storage means, the identifier reading means reads the directly corresponding identifier based on the priority read by the priority reading means.
Also, a preferable mode is one wherein, when a classification of levels of the priority given by the second storage means does not agree with that of the priority given by first storage means, the identifier reading means, based on the priority read by the priority reading means, retrieves a priority given by the second storage means and then reads an identifier corresponding to the priority obtained newly.
Also, a preferable mode is one wherein the identifier designating the communication path is a port number.
Also, a preferable mode is one wherein the identifier designating the communication path is an identifier number of a client.
Also, a preferable mode is one wherein a corresponding relationship between the priority of a service provided by the server for the distributed object saved in the first storage means and each server is able to be changed at any time.
Furthermore, a preferable mode is one wherein the change of the corresponding relationship is instructed by a client.